Is match really semantically?

[pjljvandelaar]

On https://ast-grep.github.io/advanced/core-concepts.html is written

This allows ast-grep to match patterns based on the semantic meaning of the code, not just its surface appearance.

Yet, I wonder if it is really the semantic meaning, shouldn't we find two instances when

int $VAR = 1;

is used to search this code

int i = 1;
int j = 0x1;

as both assign the value 1 to the variable, yet use a different representation thereof.

See https://ast-grep.github.io/playground.html#eyJtb2RlIjoiUGF0Y2giLCJsYW5nIjoiY3BwIiwicXVlcnkiOiJpbnQgJFZBUiA9IDE7IiwicmV3cml0ZSI6IiIsImNvbmZpZyI6InJ1bGU6XHJcbiAgICBwYXR0ZXJuOiBpbnQgJEYoJEEsJEIpOyIsInNvdXJjZSI6ImludCBpID0gMTtcbmludCBqID0gMHgxOyJ9

And similarly,
shouldn't we find two instances when

int $F();

is used to search this C++ code

int f();
int g(void);

as both declare a function without any parameters.

See https://ast-grep.github.io/playground.html#eyJtb2RlIjoiUGF0Y2giLCJsYW5nIjoiY3BwIiwicXVlcnkiOiJpbnQgJEYoKTtcclxuIiwicmV3cml0ZSI6IiIsImNvbmZpZyI6InJ1bGU6XHJcbiAgICBwYXR0ZXJuOiBpbnQgJEYoJEEsJEIpOyIsInNvdXJjZSI6ImludCBmKCk7XG5pbnQgZyh2b2lkKTsifQ==

[HerringtonDarkholme]

the semantic meaning of the code

I will change the wording of the doc. ast-grep is, as its name going, based on syntax tree. It is structural.

If you think ast-grep does not suit your need, you can also try https://semgrep.dev/ or https://about.grit.io/

[HerringtonDarkholme]

And also, if you read the whole doc, you should understand ast-grep is base on tree structure. It also has a large Textual vs Structural title on the doc.

Nearly all the paragraphs on the page is discussing the tree-structure.

Implementation wise, semantics require special logic for each language. The fact that int f() and int f(void) is equivalent is purely C/C++ semantics. They are NOT reflected in the AST.